The discovery of generalizations concerning the content and structure of phonological inventories has been a significant objective of recent work in linguistics. Such generalizations have been taken into account, explicitly or implicitly, in the formulation of phonological theories, in evaluating competing historical reconstructions, in constructing models of language change and language acquisition, and they have stimulated important linguistically-oriented phonetic research. This book reports on the work done at UCLA using a computer-accessible database containing the phonological segment inventories of a representative sample of the world's languages which is designed to provide a reliable basis for such generalizations. The project has come to be referred to by the acronym UPSID – the UCLA Phonological Segment Inventory Database.

There seem to be three types of sources for observations on phonological inventories. The type with the longest tradition is an essentially impressionistic account based on a linguist's experience of a number of languages. Statements by Trubetskoy (1939), Jakobson and Halle (1956), and Ladefoged (1971) as well as incidental remarks in the papers of numerous authors are examples of this category. Although they may be based on familiarity with a very large number of languages, there is some doubt about the scope and validity of the conclusions reached, since the list of languages represented in this experience is not given and there is no quantification attached to the statements made.

The second type consists of explicit samples of languages compiled for the purpose of a single study, such as Ferguson (1963), Greenberg (1970) and Hyman (1977) on nasals, glottalic consonants and stress respectively.